agentleFS
Sign inStart free
veraetogrex-portfolioveraetogrex-portfolio/solo-sdd/skills/context-resume/SKILL.md

context-resume

Skill0 starsChanged yesterdayMITView on GitHub

Use when returning to a project or task after a pause, a compaction or a switch between Claude and Codex, before doing any work - "continue", "let's continue", "back to the project", "check the chats", "look in the notes", "what did we decide", "where did we stop", "pick up the context", "are you up to date?", "read the history" - and when about to ask the user something they may have already said.

  • Reads credentials

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    History: Search by the task's keywords instead of reading everything.

    line 15
  2. Distinctive:

    No tool: go to the next source.

    line 21
  3. Distinctive:

    Check: Check every fact from history against files and the service.

    line 22
  4. Very specific:

    If the project rules keep handoff.md or docs/status.md, update them with the same map.

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

context-resume: a task map before the first action

The user must not carry context between sessions and agents. Everything they already said, decided or put into .env, you find yourself. The result is a short task map that shows what to do next.

Order

  1. Project rules. AGENTS.md, CLAUDE.md, solo-sdd/project.md, docs/status.md, docs/waiting.md. This is the source of truth about the process.
  2. Handoff. handoff.md in the active change's folder or at the project root, verification.md of unclosed changes, solo-sdd list --json.
  3. Current state. git status, git log -10 --oneline, the branch, open PRs, keys in .env (names only, do not print values), running processes. For a service: what is actually deployed: the commit, the container, the tables.
  4. History. Search by the task's keywords instead of reading everything. Sources by availability:
    • the user gave a live link to a chat: first read every such chat; the link title does not replace the content;
    • chat search, if the agent has it; if not, the local history files below;
    • Claude Code: ~/.claude/projects/<project-folder>/*.jsonl;
    • Codex: ~/.codex/memories/MEMORY.md, ~/.codex/sessions/<year>/<month>/;
    • a notes vault, if the project names one: start with its index note. No tool: go to the next source. Do not invent what "was in the chats".
  5. Check. Check every fact from history against files and the service. If they diverge, trust the current state and record the divergence.
  6. Map in the form below, then straight to work. If the user asked only to restore the context, stop at the map.

Map (in the user's language, without an introduction)

Goal: <what must come out, in one sentence>
Source of truth: <files, service, table I rely on>
Last confirmed state: <what was checked and where, with a date>
Decisions and constraints: <what the user already decided or forbade>
Blockers: <what we wait for and from whom> | none
Next step: <one action>
History versus fact divergences: <what is outdated> | none

Finishing work

If the project rules keep handoff.md or docs/status.md, update them with the same map. Change global memory (a notes vault, agent memory) only on a direct request.

Mistakes

ThoughtWhat to do
"I will ask the user where the key is"First .env, history, neighbouring services
"The chat said it was done"Check in the code and on the service
"I will read the whole history so as not to miss anything"Search by the task's words, history is expensive
"I will create a working copy next to it"The project has one folder, work in it
"The old note in the vault is right"A note is a hypothesis until checked
"An independent new task, but I will search the history"Do not drag foreign context into a new task

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free